At the 2026 Mobile World Congress in Shanghai, a visitor was seen interacting with a robot. In an era where large language model technology is rapidly penetrating traditional industries, a technical term from the AI domain, "Token," has evolved into what financial institutions now call the "language of credit." In March, China's National Data Administration officially named the Token "Ciyuan" in Chinese, positioning it as the "anchor of value in the intelligent age." The daily national call volume for Ciyuan has skyrocketed from 100 billion in early 2024 to over 175 trillion by June 2026, a more than thousand-fold increase in just two years. This surge in data signifies that the AI industry has fully entered the era of the "Ciyuan economy."
Recently, Guangdong Province launched its first dedicated financial product for the Ciyuan economy, the "Ciyuan Loan," in Guangzhou's Haizhu District, with three banks, including Bank of China, CITIC Bank, and Bank of Guangzhou, participating in its deployment. Bank of China was the first to finalize the initial credit line. How does the "Ciyuan Loan" alleviate capital bottlenecks for companies in the AI industry chain, and what impact will it have on the billion-yuan AI ecosystem?
The "Ciyuan Loan" is a pure credit loan product designed for asset-light AI application and tech-innovation enterprises. It uses digital metrics such as a company's monthly Token consumption data, computing power service contract values, accounts receivable from computing power operations, and API call logs as core credit credentials. AI-related enterprises have long faced a financing mismatch. These companies typically share characteristics of being asset-light, having weak financial statements, and dispersed cash flow. They lack factories, real estate, or heavy machinery for traditional collateral, with their core assets being algorithms, models, data, and code.
"A large number of AI industry chain companies are consequently rejected by traditional credit systems," said Dai Zhijie, a section chief from the Investment Promotion Bureau in Haizhu District. "These companies are at the forefront of commercialization, but traditional credit risk control language cannot clearly read their value. Banks focus on physical assets, while their assets are code. This mismatch is a structural challenge blocking the commercialization of the entire AI industry."
To address this market challenge, Haizhu District chose to take a proactive approach. It explored establishing a risk control model that aligns with the industry's logic, transforming the authentic traces of a company's sustained operations into verifiable and auditable credit references. This builds a new credit logic based on the principle of "computing power as credit." "The most critical technical support is data authenticity. API call logs and computing power usage records are inherently tamper-proof. Through simple data cleaning and aggregation, banks can convert massive raw logs into intuitive, visual operating reports, turning the confusion of 'not understanding code' into the clarity of 'reading charts'," Dai explained. In practical reviews, financial institutions also combine multi-dimensional evaluation frameworks, including computing power procurement contracts, downstream client service agreements, and technical qualifications, to cross-verify the continuity of operations and the reliability of repayment sources. This fundamentally breaks down the traditional barriers of credit risk control that heavily relied on fixed assets and physical collateral.
For AI enterprises in a rapid expansion phase, the "Ciyuan Loan" precisely fills the time gap between upfront computing power procurement and operational collections, alleviating the pain point of cash flow shortages. In Haizhu District, as one of the first to take the plunge, Guangzhou Tengyuan Digital Technology Co., Ltd. received the first 3 million yuan, three-year pure credit "Ciyuan Loan" from a Bank of China branch in Guangzhou. Tengyuan Digital is an AI tech company focused on digital marketing for three real economy sectors: home appliances and furniture, fast-moving consumer goods, and agriculture and cultural tourism. It primarily provides services like Token consumption, AI celebrity licensing, and AI tool applications. From 2024 to 2025, driven by the explosive growth of its AI marketing business, the company's revenue doubled. However, the more the business expanded, the more significant the cash flow pressure became.
"Due to industry characteristics, brand projects generally have long settlement cycles, with downstream payment cycles lagging. However, upstream costs for large model computing power Token procurement and AI licensing fees are all upfront, rigid, and prepaid," said Luo Ruichang, founder of Tengyuan Digital. The increasing capital investment driven by high Token consumption highlights a periodic working capital gap, and the old financing system simply cannot adapt to these new requirements. Previously, when seeking bank loans, the focus was on real estate and fixed asset collateral; the only option was to use the legal representative's personal property as a guarantee, resulting in low financing limits, high interest rates, and long processing cycles. The "Ciyuan Loan" completely changes the logic: banks directly recognize the daily operational digital production factors as credit credentials, precisely solving the cash flow gap challenges for businesses in the AI industry chain.
Li Liulan, Deputy General Manager of the Inclusive Finance Department at Bank of China's Guangzhou Branch, noted that the bank's initial trial credit lines have already reached 28 million yuan. The first batch of recipient enterprises is widely distributed across high-turnover, high-consumption sectors such as AI short dramas, e-commerce marketing, and AI tool development. The implementation of the "Ciyuan Loan" not only demonstrates the "timely help" of front-end capital generation through policy and financial tools but also injects financial vitality into enterprise development with real capital flows.
The deeper effect of the "Ciyuan Loan" exploration goes beyond isolated financial innovation; it lies in reshaping the entire AI industrial ecosystem. In the view of Xie Baojian, Vice Dean of the Southern China Advanced Institute of Finance at Jinan University, it successfully builds a closed-loop ecosystem where the government sets the stage to empower the Ciyuan economy, banks innovate digital finance, computing platforms provide data support, and AI service providers implement real-world applications. Xie believes the loan's launch achieves official recognition of digital productivity rights, converting intangible data into asset value and lowering the barriers to computing power and financing. For financial institutions, it breaks through the limitations of traditional credit models, opening new perspectives on AI industry risk control. For large model and computing platforms, it binds more developers and application scenarios, accelerating the large-scale consumption of computing power and Tokens. From a regional industry evolution standpoint, the loan fosters a positive cycle of financial empowerment, industrial agglomeration, and new financial demands, effectively pushing AI technology from research and development toward large-scale commercial deployment.
"The 'Ciyuan Loan' is not just a loan; it is official recognition and empowerment of the 'Ciyuan economy' and AI production capacity," said Luo Ruichang. "In the past, a Token was merely an internal pricing unit for large models. Now it has gained joint recognition from the government and banks. Stable and continuous Token consumption demonstrates that a company has real business, stable customers, and sustainable AI production capabilities. Digital productivity now formally possesses financial value."
In recent years, Haizhu District has gradually carved a development path with local characteristics by focusing on the digital economy. As of August this year, Haizhu District has gathered over 8,000 pan-AI companies, with 53 registered large models and 432 registered algorithm filings, both ranking third nationwide. A hundred-billion-yuan industrial cluster is gradually taking shape. "Agglomeration and empowerment are essentially complementary," Dai Zhijie said. In the past, the "Traffic Loan" empowered digital marketing enterprises, giving rise to a 45-billion-yuan industrial cluster. As more companies undergo digital and intelligent transformation, they will inevitably make large-scale API calls and consume Tokens, giving rise to new financing needs. Therefore, financial tools must continuously evolve alongside business growth and keep pace with industry progression.
Luo Ruichang noted that in regions with a solid AI industry foundation and data element resources, the "Ciyuan as Credit" model explored in Haizhu District has good prospects for promotion. However, achieving the leap from a benchmark demonstration to large-scale replication still faces multiple real-world challenges. "On the enterprise side, firstly, there is still a lack of unified verification of Token data and a neutral third-party audit system. Different large model platforms have inconsistent billing methods and statistical standards for Tokens, so a unified standard needs to be established. Secondly, the linkage mechanisms between industrial support, computing power supply, and financial products require continuous improvement," Luo said.
In response, the relevant authorities in Haizhu District maintain a cautious and rational attitude. Dai Zhijie emphasized that replication must not be a simple copy-paste of the product; it must be tailored to local industrial characteristics. Looking ahead, Haizhu District will continue to build a multi-dimensional risk control system, clarifying that Token consumption volume serves only as one of the credit reference standards, rather than being directly equated with collateral itself. "We will continue to have the government, platforms, service providers, and banks jointly identify promising sectors, visit enterprises, and refine products, building a new financial empowerment ecosystem that banks can understand, enterprises can satisfy, funds can be readily deployed, risks can be managed, and investment promotion can be effectively coordinated," Dai concluded.